CN107850672A

CN107850672A - System and method for accurate vehicle positioning

Info

Publication number: CN107850672A
Application number: CN201680043857.9A
Authority: CN
Inventors: C·蒂洛; P·巴尔纳德; J·赫布斯特; R·J·伦宁格尔; F·弗尔斯特灵
Original assignee: Continental Automotive GmbH
Current assignee: Continental Automotive GmbH
Priority date: 2015-08-11
Filing date: 2016-08-11
Publication date: 2018-03-27
Anticipated expiration: 2036-08-11
Also published as: EP3130945A1; US20180239032A1; JP2018525633A; EP3130945B1; WO2017025600A1; CN107850672B

Abstract

Disclose for carrying out the method and system of accurate vehicle positioning based on the supplying digital road descriptive data base comprising the 4th data set, the 4th data set includes the object-based information relevant with least one of road object, road equipment object and geographic object.The first data set based on the 4th data set is forwarded to vehicle.Vehicle is along its Path-collection environmental data, such as image, so as to generate the second data set including at least location information and object-based details.During traveling and after the second data set is generated, vehicle attempts to identify identical the driving related object and/or object datum mark and/or multigroup driving related object and/or more group objects datum marks included in the first data set in the second data set.Vehicle matches identical between the second data set and the first data set and drives related object and/or object datum mark and/or multigroup driving related object and/or more group objects datum marks, and calculates or derive the positional value of its own position.

Description

System and method for accurate vehicle positioning

Technical field

It is used to collect road data near its path and environmental data by vehicle and passes through the present invention relates to a kind of Driving associated object data is matched to the system and method for carrying out accurate vehicle positioning with the data from database.

Background technology

In order to optimize the operation of vehicle and in order to provide new feature and function, quantity is integrated with increasingly in modern vehicle Increased sensor and processor.In US 6,353,785B1, a kind of method for computer architecture in vehicle is disclosed And system.Computing architecture includes data network, processor of the data network by multiple interconnections, the environment to vehicle periphery Second group sensor that first group of sensor that condition is responded, the hardware system to vehicle are responded and comprising representative The map data base of the data of geographical feature in the geographic area of vehicle periphery.Vehicle operation Program Appliance can especially include certainly Adapt to cruise control, automatic calling for help and barrier and collision warning systems.In this embodiment, vehicle needs data network On continuous communiction, the data network there may come a time when to interrupt under the conditions of real road.Furthermore, it is necessary to accurate map datum Storehouse.

The A1 of EP 1 219 928 disclose a kind of method for being used to generate the section for numerical map.Equipped with shooting The vehicle of machine travels in a road segments.The image captured is associated with GNSS information, to produce numerical map.

In order to generate road-map, the vehicle of highly-specialised is used.They have the expensive device for being used for scanning road, Its cost may be more much higher than vehicle itself., it is necessary to carry out manual processing to acquired information after road is scanned.This compared with Expensive, labor intensity is big, and easily malfunctions.Therefore, the long update cycle can only be realized.

The content of the invention

The problem of present invention intends to solve is to improve known vehicle, database and the method for building database, institute State database and further provide sufficient information to vehicle, to realize increasingly automated driving (HAD), or autonomous driving, or nothing People drives, or other wagon control applications, or accurate vehicle positioning.

The solution of problem describes in the independent claim.Dependent claims are related to further changing for the present invention Enter.

Shown using the experiment and simulation of HAD or automatic driving vehicle, it is necessary to understand very much the environment of vehicle in detail, it is special It is not road.These data can be used for a variety of other applications, such as distance light auxiliary, speed suggestion, road camber/pit warning, economy Driving, intelligent cruise control, advanced gearshift, accelerator pedal force feedback, Intelligent Energy management, vibration damping control, advanced slow control System, cooling strategy, emergency turn auxiliary, preview ESC, intelligent parking, traffic lights auxiliary and the warning of bend speed.Therefore, it is intended that At moderate cost or low cost provides road and other newest and nearest data with driving related object.

This data are preferably based on the data obtained by the sensor in vehicle.But ideally obtain for example The road information outside the ability of sensor in long range or bend rear.This information can allow the perspective of vehicle Control.For example, if vehicle close to the crossroad at bend rear, can reduce the speed of vehicle, or before ramp Select different stalls.

In addition, these sensors may be damaged due to internal error or environmental condition.For example, the video camera of tracking road It may be damaged due to mist or heavy rain.In this case, the camera data that obtaining allows to be captured is closed with Given information The details relevant with road of connection would be desirable.So, information can be correctly explained, otherwise these information will be difficult to Identification.For example, if vehicle knows precisely the position of every of dashed centre line line (mark) and its essence from road boundary True distance, even if then camera review is very noisy and dashed centre line is difficult to detect, remaining on to be positioned.

If enough detailed, the newest and reliable information relevant with road and other driving related objects can be obtained, Can widen above-mentioned vehicle application using field or requirement to the sensor of vehicle can be reduced, vehicle can be equipped with it is less or compared with Cheap sensor.Furthermore, it is possible to driven in sensor failure or in the case of being damaged.

It is probably for navigation with reference to the general digital map for the road that the GNSS tracking of vehicle movement uses now Enough, but they are not detailed enough for automatic driving vehicle.It can be provided using professional scanning car scanning road much more Details, but this is extremely complex, time-consuming and expensive.Therefore, the update cycle may be up to the several years.Have the whole world in view, road basis is set Applying itself will not rapidly change, because road construction is expensive and time-consuming.Road between any two point protects long-time Stay, but lane specific may quickly change.Therefore, it is intended that detailed enough and accurate road information is provided in a manner of timely Or road characteristic data.

The basic conception of preferred embodiment for generating and updating accurate road characteristic data storehouse is at two step object datas Reason, wherein the first object processing steps are completed by vehicle and the second object data processing step is completed by server database.

In the first embodiment, standard production vehicle collects multiple environmental data collection using its sensor, handles them, And communicated with the accurate road characteristic data storehouse on server, with improve its collection data quality and for server database The environmental information of renewal is provided.Generally, information is based primarily upon object, and with driving related object such as road object, road equipment Object, geographic object or other object are relevant.Road object is preferably relevant with road itself.It can describe the base of road Eigen, such as width, direction, bending situation, the track quantity on each direction, lane width, surface texture.It can also be retouched State specific detail, such as kerbstone, center line, or even every line of dashed centre line, or other marks, for example, crossing or Stop line.Road equipment object and road equipment --- also referred to as street furniture --- are related.Road equipment may include for each Kind purpose is arranged on object and equipment on street and road.It may include bench, roadblock, guardrail, mailbox, telephone booth, street lamp, Traffic lights, traffic sign, bus stop, tram stop, taxi stand, public sculpture and litterbin.Geographic object can be Any stationary object, such as, but not limited to building, river, lake or mountain range.Other object can not fall within above-mentioned class Not one of in but object also related to driving.These objects can be the obstacle of tree, stone, wall or other close roads Thing.They may also comprise structure as trench, plane and can be considered for planning substitute emergency exit or collision avoid path Other structures.

Vehicle can be car, truck, motorcycle or any other device for being travelled along a path on road. Embodiment is described by taking a kind of vehicle as an example herein, but the embodiment is applied generally to a variety of vehicles.Vehicle preferably has A number of sensor, such as, but not limited to camera chain, for example be suitably adapted for capturing visible and/or infrared image Ccd video camera.It is preferably set up simple monocular-camera.Or stereo camera can be used, it, which can have, is isolated from each other Two imaging sensors of ground installation.May be present more multisensor, for example, radar sensor, laser sensor, infrared sensor, Ultrasonic transducer or other environmental sensors.Although it is preferred that sensor is orientated along travel direction, but they also can be along it Its direction is orientated, such as the side or rear of vehicle.It is further preferred that vehicle has at least one computer or multiple calculating Machine, and memory and the software for performing method described herein.Most preferably, computer, which has, is used to handle sensor The enough disposal abilities and internal memory of signal.In addition, vehicle preferably has the communicator for being used for communicating on a communication network, The communication network can be such as but not limited to the net of mobile communications network, Wi-Fi, satellite broadcasting or any other type Network, it can even is that cable network.In general, the communication between vehicle and server database passes through this communicator To establish.Vehicle can use different communication modes according to availability or cost, for example on the way use mobile communications network And Wi-Fi is used near focus or at home.

The accurate road characteristic data storehouse of also known as server database (if referring to not otherwise) is preferred Ground by service provider provide and preferably it is in trust/be arranged in multiple main frames system.It can be central database, but Preferably it is divided into the multiple subdata bases for example separated by geographic area.Preferably, data base administrator and service Device database connects.Most preferably, data base administrator includes more sub- management programs being connected with subdata base.It is preferred that Ground, server database have suitable network connection for example internet connection with vehicle communication.In an alternative em bodiment, with The communication of vehicle can perform via the rear end of third party's communication system such as vehicular manufacturer.Bandwidth should be sufficiently large, with simultaneously With a large amount of vehicle communications.Preferably, communicate and compressed and/or encrypted.Preferably, server database includes physical storage of data Device --- it can be located at particular server place, the shared infrastructure based on cloud, the combination of each approach or similar embodiment In --- and for managing the software of data storage and data flow.Here, term server database is used to refer to all wide Related system in justice.This may include that computer system, physical storage or shared infrastructure based on cloud, database are drawn Hold up, data base administrator and the other software for the data in processing data storehouse or similar embodiment.

Preferably, vehicle obtains road characteristic data before automatic Pilot and stored it in vehicle database. Initial data set can be provided by vehicular manufacturer in the manufacture of vehicle or initialization procedure, by data storage medium such as CD, DVD, hard disk, mobile memory provide, or by being downloaded from server database to provide.Because initial data set may compare Greatly, it is advantageous to available for the high-speed data connection that initial download is carried out from server database.Preferably, primary data The region or area that collection covering vehicle may be run wherein in the near future.After initial data set has been received, vehicle Further renewal may be needed.These renewals can provide and the extension or change in the object changed and/or the region previously covered Change relevant fresh information.Vehicle from server database and/or can obtain the feelings of express network connection with specific interval The extension covered under condition and/or in needs --- planning new route or be even probably this feelings during driving Condition --- in the case of ask these renewals.Server database can also trigger the data transmission to vehicle based on its cognition, Vehicle may need the data.This data are preferably the data updated and/or may include the tight of the warning of hazardous road situation Anxious data.

In order to provide initial or current data set to vehicle, server database is according to being stored in server database At least one first data set of the 4th data set generation.Request or the server of at least one first data set and vehicle The triggering of database is related.Preferably, the target zone of the first data set covering vehicle.This target zone can make in vehicle By server database or vehicle modification during.Preferably, the path of vehicle traveling is a part for its target zone.It is excellent Selection of land, server database have been stored with the information relevant with the state of vehicle database.This information may include to be stored in car The backup of the first data set in or at least one identifier for identifying the first data set being stored in vehicle.The shape State information updates after the reception of the first data set is preferably had determined that in vehicle.By using the status information, the first data Collection can only include amended data set and/or new data set.Data set, the vehicle that vehicle does not have, which may be present, to be had but more The data set or server database that new data set, vehicle have but be retransmitted wish the data set that vehicle is deleted.If by Server database triggers, then at least one first data set can include the data updated and/or alarm data and/or should When the data verified by vehicle.Preferably, each first data set includes at least one location information with detailed based on object Information.Location information specifies the detailed applicable position of object-based information, region or area.This includes, data set It may include one place information and multiple object-based information preferably related to same or adjacent place.It is object-based Information may include relevant detailed with least one driving related object such as road object, road equipment object and geographic object Information.

Preferably, vehicle is supplied to by server database and is supplied to the data set of server database to include from vehicle Object-based 3D vector formats.Most preferably, first, second, third and fourth data set includes this vector format.It is this Vector format can be Vector Graphics, the normal vector being preferably such as used in computer game or CAD draughtsmanships Picture format or derivatives thereof.Preferably, the boundary line of the detailed object of road such as road, the graticule of road, intersection and go out Mouth and street furniture are represented with 3D Vector Graphics.Due to Vector Graphics, it is only necessary in vehicle and server database Between transmit low volume data back and forth.In addition, Vector Graphics allow to render road characteristic data by different way, to represent not With viewing angle or sensor pattern, such as video camera formula or radar type.

Preferably, auxiliary information is supplied to vehicle by server database.It is this auxiliary Supplementary information can be confidence levels/horizontal or other information relevant with confidence level of server database generation --- referred to here as For confidence levels --- and importance information, and urgency or statistical information such as average value and standard deviation.It can be directed to each Single data set or a group data set provide the auxiliary information.

After the first data set has generated, they are forwarded to vehicle and are stored in vehicle database.Vehicle can be with The data relevant with driving are derived for its drive control application, such as automatic Pilot from vehicle database.

But vehicle is not only the user of road characteristic data.They are also by analyzing the ring by their sensor collection Border data set generates road characteristic data.Under common road and riving condition, it may be preferred to which vehicle is in traveling Simultaneously, while traffic is participated in or at least in the sensor on road or while neighbouring road by them, preferably Ground is also sampled or collected to information by their imaging sensor, and periodically or periodically by sample information with Download in advance or the information of storage is compared and/or associated.Between this collection frequency or time interval --- it also may be used Referred to as sample rate --- it can be determined by the disposal ability in the sensor in vehicle, their resolution ratio and/or vehicle.Generation For periodic sampling, stochastical sampling may be selected.Also environment sampling rate adjusting can be directed to.In the case of crankcase ventilaton, with forthright Compared to may want to higher sample rate.Server database can also ask different sample rates.

The environmental data collection sampled can be recorded and stored certain time as a series of environmental data collection.This allows to make By the environmental data collection that stores of background task processing, and move back and forth stored data set, with carry out with must be in speed More complicated processing is compared in the real-time processing of upper following vehicle.The storage of a series of environmental data collection sampled allows vehicle Particularly when driving over time and/or distance and supervision object appearance.For example, traffic sign can pass through video camera From a dot on the image of wide-long shot.When close to traffic sign, the point expands and starts to show characteristic pattern, directly It can be clearly seen to it shortly before by the traffic sign in zoomed-in view.

Continuous record of the environmental data collection at least on certain time section also allows for reviewing.In the examples described above, car May be just by identify the traffic sign before traffic sign.Now can when close to traffic sign rollback number According to collection, such as image, and this traffic sign is generated from seeing at a distance as what or it are exactly in the more information of where. The data may also used as driving a part for related data, to enable the vehicle to detect traffic sign from farther distance.

Server database can transmit sample request message, and the request message is preferably specified and the information should be carried out The time of sampling and/or place.In addition, server database can ask the specially treated in special object or region.For example, it can Request carries out text or Language Processing to traffic sign.It can also ask the height of zone of ignorance --- for example, traffic sign --- Image in different resolution so that server database itself can carry out other processing, such as foreign language text identification.Particular request can Preferably it is addressed to particular vehicle, the particular vehicle can have particular device, such as stereo camera or laser scanner, or It can move in a particular area.Server database can also upload particular procedure instruction or specific program code.These special places Reason request can be used as implicit instructions or is contained in by parameter or data at least one first data set.For example, low confidence level Rescaning for object will not be triggered, to obtain better information and therefore raising confidence levels.

Sampling may be based on confidence and/or importance and/or urgency and be triggered.If first in vehicle database Data set shows in locality than relatively low confidence levels or high importance and/or urgency, then it is preferably most of or Information will be collected and renewal is forwarded into clothes by their sensor by passing almost entirely through or rest on the vehicle in the place Business device database.Elapse over time, increasing data are collected, and this causes higher confidence levels.Due to may not Need to rescan object with higher confidence levels, so the number of objects rescaned is reduced over time.Therefore, vehicle and Network traffics and required disposal ability in server database can decline.

In addition, if a large amount of vehicles just update in simultaneous transmission, priority scheme may be present to better profit from comparing just Suitable communication network.If for example, data of the confidence levels of data set than the transmission improvement of relatively low and sensor, vehicle can Record this data.If confidence levels is very high, a small number of difference can be only recorded.In this case, server database It can further improve confidence levels.If data have highest confidence levels, vehicle will not be recorded in the region more More data.On the contrary, it can generate at least one confirmation data set, or will be with process in the case where not yet identifying any difference The relevant confirmation message of specific road section is sent to server database.So, the traffic and thus communications cost can be with data Gradual growth confidence level and reduce.

Disposal ability in some vehicles may be not enough at least with high frame per second --- its for the vehicle run at high speed and Speech is probably enough --- the biography to the data provided by vehicle database and for complete vehicle running path by vehicle The data that sensor obtains are compared in real time.Carried out with the 120km/h vehicles travelled with 33.3m/s.In order to obtain about 1m's A series of spatial resolution of environmental data collection, it is per second to handle 33.3 environmental data collection.Related to other driving appoints It is engaged in and is about to the burden that vehicle database information may aggravate the computer of most of vehicle compared with environmental data collection, it is special It is not because single environmental data collection may include multiple sensors such as video camera, radar sensor, infrared or RF sensors and such as The bulk information of the more multisensor of other environmental sensors.Therefore, this is than preferably in the larger, discontinuous of vehicle route Section in complete, as that can be handled by each vehicle.The length of vehicle route section can be at several meters to hundreds of meters of scope It is interior, and then several meters to hundreds of meters of gap, the gap can further depend on the speed of vehicle or currently may be used each other The disposal ability of acquisition.In a further embodiment, the assessment of different information can independently be carried out.For example, visible video camera Image can be assessed in real time, and the image of laser scanning can be assessed in the time below.In a further embodiment, information is certain The assessment of rank is close to and carried out in real time, and with the following vehicle in speed, and the more detailed assessment of information can detected only Carried out in the case of difference.

Vehicle is further assessed by the road environment data set of sensor collection what is all or at least selected as described above To generate at least one second data set, each second data set comprises at least location information and set with road object, road Apply the relevant object-based details of at least one of object and geographic object.In order to generate this second data set, Assess the environmental data collection of storage.Preferably, vehicle is only by the environmental data collection of storage or by by the environmental data of storage Collection identifies driving related object with the information association from least one first data set being stored in vehicle database.Such as For example bad visual field of fruit, bad weather conditions may cause object to be difficult to, or if object is at least partly hidden, then This association is probably useful.Preferably, advanced technology is used to improve object factory precision by vehicle, such as later will be more detailed The statistical estimation of the object carefully described or packet.Second data set represents current visual angle of the vehicle to driving related object.It is raw It is that the first object data completed by vehicle is handled at least one second data set and/or at least one 3rd data set of generation Step.

After the second data set is generated, vehicle generates the 3rd data set, and the 3rd data set includes and is stored in vehicle number According to the difference between the object-based information of the second data set in storehouse and the object-based information of the first data set.It is additional Ground or alternatively, at least one 3rd data set may include the second data set, without the difference calculated.In addition, the 3rd data Collection can include component environment data set, complete environmental data collection, the sequence of environmental data collection and from the other of sensor Auxiliary information and vehicle-state and intermediate processing results.Preferably, the overlay area of the 3rd data set such as server data place As request or as limited as determined by vehicle.The selection described in detail above.It is also preferred that when second When minimum difference at least be present between the object-based information of data set and the object-based information of the first data set, unless The each data set of server database clear request, otherwise only generates this 3rd data set.As not generating specific data The alternative solution of collection, data set can still generate but be marked as not transmitting to server database.If vehicle has identified Go out significant data and generate at least one second data set without the first data set corresponding to generation, then may be assumed that vehicle has identified Go out new object.Preferably, the 3rd data set is generated to these new objects all the time.3rd data set can be included and can provided further The assistance data of the information such as confidence levels of vehicle generation.The confidence levels of this vehicle generation can by vehicle setting based on vehicle Standby, travel speed, pendulous frequency or image are determined based on signal to noise ratio or any other measurement parameter.More parameters may be present To influence the confidence levels of vehicle generation, such as weather conditions, illumination condition, shade, blind area and other conditions.

Finally, the 3rd data set is forwarded to server database by vehicle.Being communicated with server database can be according to communication The availability of network is carried out.Therefore, communication can enter in vehicle parking or when traffic light is stopped and can obtain Wi-Fi OK.Important and/or urgent data can exchange via mobile communications network all the time, and other data can obtain later Exchanged during Wi-Fi.Due to the 3rd database mainly include drive related object object data, so they it is compact and Required size is smaller than the image from road or object shooting.Which greatly simplifies communicate and alleviate network load.Only very In the case of special, bigger view data can be sent to server database.For example, if traffic sign is unreadable , then the image of the traffic sign can be forwarded to server database and be assessed for further.

Above-mentioned steps can perform successively, so as to substep or handle a number of data by handling single data set Collection.For example, the data of certain section of vehicle route can be collected and the step-by-step processing after vehicle has been subjected to the path-segments. Conversely, data processing can be carried out continuously while vehicle passes through the particular section in its path.In addition, data can be continuous It is forwarded to server database or is transmitted.Combination is also possible.For example, if vehicle detection goes out important object, When vehicle passes through the particular section in its path, can be carried out while all associated sensor datas are recorded continuous real When handle, so as to allow to review as described above.

Preferably, server data library storage is to the selection of the 3rd data set received.Or server data stock Store up the 3rd data set of all receptions.Preferably, the data set of big flow control three for every road is stored in server database In.This can be the historical record for the result for showing previous the 3rd data set from different vehicle for the road.Clothes Business device database at least has the 4th and the 3rd data set that can be used for further assessing.Server database can be one Partial data base administrator can look back the 3rd data set from vehicle receiver relevant with the difference of the 4th data set.Service Device database is attempted the 3rd data set being fitted to the 4th data set.Generally, the location information of the object of the 3rd data set should Close to the 4th data set.The deviation for the maximum GNSS tolerances for about limiting maximum allowance window may be present.Therefore, server data Storehouse must only compare the data set in the maximum allowance window.The processing performed by server database is to be used to generate and update The second object data processing step in accurate road characteristic data storehouse.It can also be completed by following steps and method.

Database preferably generates the object datum mark for object.This object datum mark from now on can be used for match/ With (align) object or more group objects.Object datum mark is preferably allocated to what can be identified easily and with higher precision A part or feature for object.It can be the geometric center of traffic sign or the particular lamp of traffic lights.

Preferably, server database from the information of vehicle receiver to carrying out statistical estimation.If a large amount of vehicle report phases Same difference, then there is a possibility that certain real road has changed compared with server database.Therefore, server count may be updated According to storehouse.

One very important aspect is that the object factory precision of the 3rd data set is improved when generating four data sets.Example Such as, the object modeled in the 3rd data set carried out by vehicle may not be optimal for the 4th data set, and can Changed when generating four data sets by server database.In addition, error, and not all mistake easily occur for GNSS location information Difference all has normal distribution, error can be handled by averaging.For example, the first quantity from the first vehicle The 3rd data set relative to the second quantity from the second vehicle the 3rd data set can have specific position skew.Or Person is, it is necessary to handle the 3rd data caused by vehicle sensors lack calibration and standardization --- such as optical skew --- Concentrate retained error.In order to improve object factory precision, multiple data sets can be assessed and/or matched --- the preferably the 3rd number According to collection --- information, especially positional information.

Preferred embodiment is related to following accurate proportioning vehicle route data with same with road by matching and matching Part related particular data set improves the method for collected data.Preferably, server database is from multiple vehicles Receive the 3rd data of at least one environmental data collection related to each section of vehicle route that report is determined based on vehicle Collection.These environmental data collection relate preferably to drive at least one of related object, road equipment and geographic object.For life Into these the 3rd data sets, object data processing is performed by following steps in vehicle：Assess at least one environment number Associated object data is driven according to collection and based on assessment result and/or component environment data and/or intermediate processing results generation extremely Few second data set, and by least one second data set for driving associated object data and it is stored in vehicle data At least one first data set of driving associated object data in storehouse is compared.

It is each based at least one 3rd data set, server database identification overlaps each other at least in part and road A part of relevant multiple first vehicle route sections.These can be pre-selected based on the location information relevant with data set Data set, the location information can be GNSS location information.It is followed by based at least one in each vehicle route section In location information pair included in 3rd data set multiple first vehicle route sections related to a same part for road The first thick proportioning step that 3rd data set is carried out.In order to accurately match, identified at least two vehicle route sections multiple Object datum mark and/or related object.Datum mark and/or related object be grouped into so that：Multiple first vehicle route sections In at least two vehicle route sections include at least one set of identical object datum mark and/or related object.Finally, generation or Calculate at least two vehicles for matching and/or matching the multiple first vehicle route sections related to a same part for road The proportioning function of at least one set of object datum mark and/or related object in path-segments.Identical proportioning function is at least Suitable for the selection of another 3rd data set of the vehicle route section of identical at least two, phase is driven to generate or update to include Close the 4th data set of object data.This proportioning function may include linear translation and/or rotation or even more complicated conversion Any one of.

It may further be preferable that server database uses the 4th data set from server database and the 3rd number According to collection as the basis for matching.

Generate and application proportioning function after, can by preferably calculate proportioning the 3rd data set average value and/ Or standard deviation statistically handles the 3rd data set.

Further preferably carry out other proportioning and/or by using the datum mark at intersection and/or from satellite image The position known and/or the geographic reference points of fixation match.

According to further embodiment, server database, which has, performs the accurate proportioning car related to a same part for road Means needed for the method for path data.Preferably, this method is by the calculating with the computer program in memory Machine performs.

Obtainable first and second data sets in vehicle can be used to perform same procedure to improve the second data set for vehicle Precision.It is preferred that the vehicle route related to a same part for road is performed in server database rather than in vehicle The accurate proportioning of data, provided because server database has considerably higher disposal ability and may have access to by a large amount of vehicles Mass data.

Even if so, vehicle also can perform packet for another purpose.Point of object datum mark and/or related object Group can be beneficial to accurate vehicle positioning.A large amount of objects with relative position benchmark, such as center line line may be present along road. Due to substantially looking similar line in the presence of many, it is difficult to make a distinction and be taken along the vertical of road between each line To the absolute position benchmark in direction.Anyway, even if so, line can also be used for determining that vehicle is orthogonal relative to road boundary In the position of road.Indulged to verify and correct GNSS location information and/or vehicle relative to the object of road and along road To being accurately positioned, it is necessary to more position references for direction.These position references can be unique object or pair seldom occurred As.These benchmark can be the traffic sign or distinctive mark of on road or both sides of the road.But special object may not very frequently Occur numerously and be typically not enough to be used for accurate and continuous vehicle location.Therefore, it is further preferred that multiple object quilts It is grouped together.Even if the object of each type in group may repeatedly occur along road and not be easy to differentiate, this Plant the group of object and especially its cluster formed in the 3 d space is also likely to be unique for very long one section of road. Example is the cluster of two line and the roofbolt formation of guardrail.Their positions relative to each other are not yet planned and therefore non- Often it is likely to be unique.Even if the really unique single object of None- identified, the packet of object or object datum mark also can Determine exact position of the vehicle on road.The exact position of vehicle can be expressed as inclined with the position that is provided by GNSS system Move.It can also be expressed by the positional value in the world locations of the first data set, it is meant that with being wrapped in the first data set The driving related object and/or multigroup driving related object contained is related.

Following methods describe a preferred embodiment of accurate vehicle positioning.Preferably, the identification pair in the first data set As datum mark.In motion, vehicle generates the second data set according to collected environmental data collection.In the second data set, car Attempt same object included in the first data set of identification and/or object datum mark and preferably more group objects and/or right As datum mark.Second data set of vehicle is generally by only comprising the selection to the data included in the first data set.By with Than the same object between the second data set and the first data set or object datum mark, preferably more group objects and/or object base On schedule, vehicle can derive the pinpoint positional value of its own position and the position on road.

According to further embodiment, first, second, third and fourth data set may have further information, example Such as confidence levels, so as to provide the information relevant with the confidence level of information or reliability.If many vehicles report identical difference, Then confidence levels is higher.If only one vehicle or a small number of vehicles report the difference, confidence levels is relatively low.Confidence levels also may be used Confidence levels depending on vehicle generation.Therefore, the data set of the confidence levels generated with higher vehicle can be with higher Weight coefficient.Another coefficient can be the sensor configuration grade of vehicle.Preferable vehicle is configured it is determined that during confidence levels There can be the weight higher than configuring relatively low vehicle.Generally, two different confidence levels values --- it is such as high-level and rudimentary Not --- it is sufficient that.The confidence levels with higher amount is may want to, for example allows three ranks distinguished in more detail Or nine ranks.If vehicle reports obvious fresh information first, confidence levels may be relatively low, but this be probably it is important or Urgent information.Therefore, data base administrator preferably generates at least other mark, its may indicate that information importance and/ Or emergency.Or confidence levels can be set to high value, with the higher significant or emergency of configured information.It is for example, dynamic The speed limit changed to state is probably emergence message, because the vehicle drive more than speed limit will be reported.And the change of road sign Changing may be less urgent, because the process for changing road sign expends a few hours or a couple of days, but it is also likely to be an important letter Breath.The construction of new building beside road is both inessential or not urgent.But obtain the information relevant with building and can improve and determine Position precision.Generally, it is sufficient that two different important and/or emergency level --- such as high-level and low levels ---.May Wish the rank with higher amount, for example allow three ranks or nine ranks distinguished in more detail.

Generally occur if server database understands some changes in locality, such as may be due in locality Velocity of electrons shows and frequently velocity variations be present, then can assign higher confidence levels.In addition, vehicle preferably also provides not The confidence levels generated with the vehicle of information.Server database can also be based on the cognition relevant with vehicle or environment --- such as Sensor quality (such as it is low resolution or or high-resolution camera) and data quality is in itself --- to change confidence Rank.In another example, high-resolution camera will be provided than the low resolution during snowstorm in the picture rich in detail of fine day Image wants high confidence levels.

Generally, the data such as numerical map, satellite image or transport information of more data provider can be used to improve number According to the positional precision and/or supplementary data set of collection.

Further embodiment is related to a kind of mitigates total image processing load in vehicle and such as using the first data set Expand the method for the increasingly automated use range for driving (HAD) and the application of other wagon controls towards severe weather conditions.

Such as HAD wagon control application depends critically upon the processing of the view data from imaging sensor.At image Reason needs the big energy-handling capability in vehicle.Each type of image processor generally has limited use range and its specific Pros and Cons.For example, video camera provides high-resolution colour picture, but there is low-dynamic range.They are for including light and shade The scene in region is difficult.Video camera can not also work well in mist or in the rain.On the other hand, radar work under these conditions Make well, but the low resolution monochrome image for including reflectivity information is only provided.In order to compensate the inferior position of dissimilar sensor, Various sensor types and multiple images sensor are combined in vehicle to expand HAD and the application of other wagon controls use Scope.Cause a large amount of required disposal abilities to the demand for assessing and merging the view data from multiple images sensor.No Cross, the use range of HAD and the application of other wagon controls is fairly limited, because while having the sensor device of costliness, severe day Gas bar part also still brings challenges.

Needed as described above, the method for generating and updating accurate transportation database adds image procossing other in vehicle Ask, so as to processing environment data set, generate the second data set, by the second data set compared with the first data set, and it is raw Into the 3rd data set for being sent to server database.By compared with the first data set, being essence by environmental data collection True vehicle location adds bigger disposal ability.At least the latter must be all the time with car speed real-time working.

Preferably, the present embodiment includes processing and use two stages of the data set in vehicle.In the first phase, take Business device database collects the 3rd data set of the section of vehicle route and assesses them is further used as the first data set with generation The 4th basic data set, so as to establish the confidence level in the 4th data set.In the first stage, the first data set can be with Vehicle is supplied to by server database, and it is related right to driving in vehicle side to guide for generating the second data set by vehicle The processing of the modeling method of elephant., but it is preferred that in the first stage of the confidence level for lacking object data, confidence levels is low The first data set by vehicle be used for environmental data collection processing.

In second stage, high confidence level has been reached by least one in the first data set of the 4th data set generation Not, because server database collected sufficient amount of 3rd data set so as to assess they and generate have precise results and 4th data set of high confidence levels.Server database is by the 4th the first data set of data set generation and sends them to Vehicle.The first data set with high confidence levels can be used for processing environment data set and wagon control application.Usage type can Determined by confidence levels value.For example, the first data set with very high confidence levels can be used for increasingly automated driving in mist Sail, and the function the confidence levels of corresponding first data set be not it is very high in the case of fail.On the other hand, in the first number It is higher according to the confidence levels of concentration but in the case of being not extra high, it can still allow to drive in the rain small, wherein video camera " is seen To " more.Environment and/or riving condition adjustment limiting value can be directed to.

First stage described further below and second stage.

Preferably, in the first stage of confidence level is established, by being sampled simultaneously to the environmental data collection for one section of road The sequence for the environmental data collection that processed offline obtains limits the processing load on vehicle, it means that processed by skipping over Section between section is without with the speed follower vehicle of vehicle.The constraint for server database collect the 3rd data set and Speech is not shortcoming.The collection of 3rd data set may rely on many for the 3rd data set for providing each section for its path Vehicle.Without providing the complete information relevant with its fullpath by a vehicle.Each vehicle can sample and storage environment number Carry out environmental data collection described in processed offline according to the sequence of collection and the maximum processing capability having with it and the maximum duration needed for it Sequence.If the vehicle miss out the part in the path between each section in the path of its processing, these gaps will be by Other vehicle fillings.

Generate the second data set and/or information for wagon control and/or the information for vehicle location and/or storage The sequence of environmental data collection and the method for processed offline have various option of operation, have essentially described these operation choosings above , concluded below.Any one or more in these option of operation may be selected to supplement processed offline：

The scaling option：Selection amplified close to object to be processed and for good recognition result with clearly Show its environmental data collection.Farther environmental data collection can be used for multiple step format triangulation of the object relative to other objects And positioning.This obtains good identification using lean Processing Algorithm and is accurately positioned result.

Feature option：Selection is concentrated close to the environmental data collection of object of interest in each environmental data.Due to only near Field is interested, wherein identification and relative measurement work are good and/or can use lean Processing Algorithm.

Move back and forth option：Move back and forth environmental data collection so as to from closely or with different view and distance and therefore Object of interest is determined with relatively low disposal ability.

Sequence option：A series of environmental data collection is run successively by different algorithm groups.Not necessarily like real-time processing All objects are handled and identify as it is required that simultaneously.This improves while the requirement to disposal ability is substantially reduced and answered The possibility of disposed of in its entirety.

Look back option：Perform the retroversion processing from nearest environmental data collection to past environmental data collection.With for example The performed real-time processing of HAD control applications on the contrary, need not see forward from nearest data set its far field and it is therefore seen that In the future that vehicle will face.In addition, this causes lean algorithm and required reduction process ability.

Generally, the fact that can carry out processed offline to the sequence of environmental data collection in the first stage, causes use The good result of lean algorithm and the low burden of required disposal ability.

Will be in the case where treatment conditions be good it is also preferred that generating the 3rd data set owner in the first stage --- Mean under good weather and good illumination condition --- complete.The 3rd data set need not be generated under severe conditions to draw Rise even for good result, also can mainly only rely upon video camera and be used as environmental data collection and required Reduction process ability imaging sensor.Due to the processing of environmental data collection can complete under good conditions and this first It need not be handled in real time in stage, so obtainable disposal ability can be used for generating the 3rd data set, the 3rd data set With it is many driving related objects, complete filling of data and in the first stage be directed to for vehicle route section it is good Precise results needed for 3rd data set.

In second stage, high confidence level is reached from least one in the first data set of the 4th data set generation Not.The first other data set with low confidence levels may be still suffered from.The first data set with high confidence levels only needs Detected and confirmed by the vehicle Jing Guo corresponding object.Preferably, vehicle actively confirms the first data in the 3rd data set The data of concentration, until having reached very high confidence levels.Preferably, once the data in the first data set have reached very high Confidence levels, vehicle just their paths for only being travelled in the 3rd data set of flagship report.By not reporting the feelings of difference Their driving path, data of the vehicle indirect acknowledgment along first data set in the path are reported under condition.Which reduce the traffic. It is also preferred that the run routing information of vehicle is neutralized so that it can not be traced back to driver and improve privacy concern. Preferably, how many vehicle server database monitors and when directly or indirectly confirms data in the first data set Which part.If there is no enough confirmations, preferably server database will be corresponding in the 4th data set The confidence levels of data changes to corresponding relatively low value.

If vehicle only can actively or a part of the indirect acknowledgment along first data set in its path or report difference take Business device database will need to track which part of the data which vehicle confirms the first database.Therefore, server data Storehouse will need more specifically to understand which vehicle moved up at what time in which path it is dynamic.This may bring privacy concern.For Such case is avoided, each vehicle is preferably able to confirm all first data sets or detection difference along its path.It is this to want Asking means that requirement is at least handled all environmental data collection in real time to a certain extent.

In order to realize the low burden to disposal ability, it is preferred that be used for the first data set since high confidence levels Support the processing of environment data set.The support is provided to driving related object of the generation for the second data set, and The support can be provided by two kinds of entirely different methods：

Method 1：As described above, the first data set includes the object factory of the driving related object on airway road.It can be used Same object in quick detection and inspection environment data set.However, therefore, before it can carry out this comparison, first Environmental data collection at least must be managed with certain object level.For example, if stop sign is covered by snow part, substantially not It may identify it as the object compared with the object data of the first data set.

Method 2：As described above, the first data set preferably includes Vector Graphics, it allows by different way simultaneously Image is rendered as with different view.The each image concentrated for environmental data, can be presented airway road or part thereof of phase Answer image, it illustrates airway road in expected place and block respective sensor " viewing " airway road driving it is related right The view of elephant.This allows the comparison to image level., can be with for example, in order to compare the image from camera sensor Respective image is presented in a manner of people also so sees and from the visual angle of corresponding video camera.For the image from radar sensor, It can utilize reflectivity and run time information that respective image is presented from the first data set in a manner of radar sensor sees the world. The environmental data collection having been positioned in image level is allowed faster and with higher level really compared with the first data set Fixed degree and/or lower disposal ability obtain correct conclusion.The snow-covered stop sign in part now can be by by position It is in compared with the image on the environmental data collection in image level with the airway road under good condition and by identification The shape of the stop sign in correct place is identified, and can be some punctations, wherein the part indicated is not Covered by snow.Method 2 contributes to utilize the image from the first data set generation real as the ground compared with a kind of be used for people Condition experience carries out the similar mode of image procossing to carry out Object identifying.

Method 1 and method 2 can also use in hybrid form.For example, first, sent out for detailed and quick detection Before the raw comparison to image level, the airway road driven in related object the first data set of inspection to be detected can be directed to Object factory.The processing can also jump with towards good result back and forth between the graphical representation on object factory and airway road Iteration.

So, the first data set can be used for the generation for supporting the second data set.It reduce required disposal ability, and Allow processing seamlessly rapid following vehicle, to confirm to server database along the complete of the first data set of vehicle route The difference between the first data set and the second data set in whole content, or the 3rd data set of report.The processing of environmental data collection At least part that task is transferred in second stage from the segment processing in the first stage is handled in real time.

In addition in the 3rd centrally updated server database of data, the first data set can also be used to support environmental data The assessment of collection and with the first data set be compared to accurate vehicle positioning, this is also a real-time task.Can not have With the help of first data set it is quick and/or with relatively low disposal ability identification and comparison other and/or object datum mark and/or More group objects and/or more group objects datum marks.

In an identical manner, the first data set can also support what is completed by HAD functions or the application of other wagon controls The real-time processing of environmental data collection.It reduce the renewal institute in parallel for carrying out HAD, accurate vehicle positioning and accurate transportation database The disposal ability needed.

The support of the image procossing of the first data set carried out particularly by method 2 also extends towards severe weather conditions HAD and the use range of other wagon controls application.The side handled for the environmental data for supporting to carry out by the first data set Method 1 and method 2 can be used for wagon control application.Foregoing disclose the example of the snow-covered stop sign in part.It is another Example is the automatic Pilot at night.Video camera " can only see " the so remote distance of headlamp irradiation light.Radar can be seen It is farther, but road sign " can not be seen ".As long as video camera starts to identify road sign, the first number at the edge of its visible range According to collection can by provide road sign by positioned at where or go expection whither come " seeing " vehicle more in front of road Mark.These are only a few examples how the first data set helps HAD and the various possibilities of other wagon controls application.

Obviously, the order of some steps in these steps can be changed, on condition that it is any above the step of provide it is latter Data needed for step.

Above-described embodiment, which is related to, to be used to include driving correlation by using vehicle, server database operation, processing, storage The data set of road object data and with vehicle communication come generate and update include drive related object accurate link characteristics number Related data, this data of assessment and the dress to be communicated with server database are driven according to the method in storehouse, and with being used to collect The vehicle put.Preferably, this method is performed by the computer with the computer program in memory or multiple computers.

Brief description of the drawings

Below with reference to accompanying drawing illustrating the example of embodiment without illustrating by way of limiting overall invention thought The present invention.

Fig. 1 shows the general scene of the automatic driving vehicle travelled on road.

Fig. 2 shows basic vehicle component.

Fig. 3 shows that accurate vehicle position is detected.

Fig. 4 shows that road characteristic data updates.

Fig. 5 shows that Essential Environment data set obtains.

Fig. 6 a, 6b and 6c are shown by the different images of the vehicle shooting of movement on roads.

Fig. 7 shows that complete environment data set obtains.

Fig. 8 shows the example of object packet.

Fig. 9 a and 9b are shown by being grouped object the first example for being matched.

Figure 10 a and 10b are shown by being grouped object the second example for being matched.

Figure 11 shows the basic procedure of Data Collection.

Figure 12 accurately matches the detail flowchart of the vehicle route data related to a same part for road by vehicle.

Figure 13 accurately matches the detailed of the vehicle route data related to a same part for road by server database Flow chart.

Figure 14 shows pinpoint detail flowchart.

Figure 15 shows to establish basic 2 stage method of transportation database.

Figure 16 is shown specifically the stage 1.

Figure 17 is shown specifically the stage 2.

Figure 18 shows the image of exemplary roadway.

Figure 19 shows Figure 18 driving related object.

Figure 20 shows the image presented from object.

Embodiment

In fig. 1 it is shown that the general scene of the automatic driving vehicle travelled on road.Road network 100 includes a plurality of Road 101.Each road has a particular characteristics, for example width, direction, bending situation, the track quantity on each direction, track are wide Degree or surface texture.There may be more specific details, such as kerbstone, center line, or even the wall scroll line of dashed centre line, or Other marks, such as crossing.Near roads have the geographic object 140 related to driving.Geographic object can be any quiet Only object, such as, but not limited to building, tree, river, lake or mountain range.Additionally, there are multiple road equipment objects 150, its It may include the object and equipment being arranged on for various purposes on street and road, such as bench, roadblock, guardrail, mailbox, phone Booth, street lamp, traffic lights, traffic sign, bus stop, tram stop, taxi stand, public sculpture and litterbin.It may be present Other object, it does not fall within one of above-mentioned classification but related to driving yet.These objects can be greenbelt, tree, stone, wall The barrier of wall or other close roads.They may also comprise structure such as trench, plane and can be considered for planning and substitute tightly Escape vent or collision avoid the other structures in path.

In addition, service provider 120 preferably has at least one database hub being connected with server database 122 121.In order to be communicated between service provider 120 and vehicle 110, there is provided there is communication network 130.This communication network can Be mobile communications network, Wi-Fi or any other types network.Preferably, the network internet protocol-based.Embodiment party Formula relates generally to vehicle, and it can be car, truck, motorcycle or any other device for being travelled on road. For simplicity, car is shown in figure.

In fig. 2 it is shown that basic vehicle component 200.Preferably, vehicle includes environmental sensor 210, vehicle-state passes At least one of sensor 220, position sensor 230 and communication system 240.These sensors are preferably by least one logical Letter bus 259 is connected with processor system 250.The communication bus may include single bus or multiple buses.Processor system 250 It may include multiple individually processors 251, it is preferably connected to each other by bus 252.In addition, the multiple individually processing Device is preferably connected with vehicle database or memory 253.

Environmental sensor 210 collects the information relevant with the environment of vehicle.They may include to may be adapted to capture it is visible and/or The camera chain of infrared image, such as ccd video camera.It is preferably set up simple monocular-camera.Or solid can be used Video camera, it can have two imaging sensors of installation of turning up the soil that are isolated from each other.More multisensor may be present, for example, it is at least one Radar sensor or at least one laser sensor or at least one RF channels sensor or at least one infrared sensor are another A kind of environmental sensor.Sensor can be used for scanning and detecting external object.

State sensor 220 collects the information relevant with vehicle and its internal state.The detectable traveling shape of these sensors State, travel speed and steering direction.

Position sensor 230 collects the information relevant with the position of vehicle.These position sensors preferably include GNSS System.Any alignment system, such as GPS, GLONASS or Galileo can be used.Here, table is carried out using only term GPS or GNSS Show all alignment systems.It also can exist for determining gyroscope, yaw angle transducer and/or the accelerometer of the motion of vehicle.Also It is preferred that with the range sensor invested on wheel, such as turn-sensitive device, it can be used for more multisystem, such as measuring row The anti-locking system or anti-lock braking system (ABS) of distance are sailed, and/or for detecting the rotation direction sensor of travel direction.It may be present More multisensor, such as accurately detecting the altimeter or other height above sea level or obliquity sensor of altitude change.Position sensing Device 230 can interact or using the signal from state sensor 220 with state sensor 220 at least in part.

Communication system 240 is used for the equipment communication with outside vehicle.Preferably, communication system is based on communication network, and this is logical Communication network can be mobile communications network, the network of Wi-Fi or any other types.Different communication network can be used in communication system And/or communication protocol, depending on their availability.For example, in the stand of vehicle, Wi-Fi can be obtained.Therefore, at this A little positions, Wi-Fi are probably preferable communication network.In addition, Wi-Fi can at the parting of the ways, traffic lights are nearby and with low The section of traffic speed or regular congestion nearby obtains.If unavailable Wi-Fi, other communication networks can be led to as mobile Communication network is used to communicate.In addition, communication system 240 may have the buffering of delay communication, until suitable network can be obtained.Example Such as, when vehicle has reached parking place and can obtain Wi-Fi, the road characteristic data distributed during traveling can be turned Issue service provider.

Processor system 250 can be the processor system that can be obtained in a large amount of vehicles.In modern vehicle, make Different task, such as engine management, Driving control, automatic Pilot person auxiliary, navigation and amusement are controlled with many processors. Generally, very big processor power/disposal ability can have been obtained.This obtainable processor or other processor Available for performing task described herein.Processor is preferably connected by bus system, and the bus system may include multiple differences Bus, such as CAN, MOST, Ethernet or FlexRay.

In fig. 3 it is shown that the elementary stream of accurate vehicle position detection.Can be by alignment system as GNSS is realized Positional precision is not enough to be used for automatic driving vehicle.10m normal place tolerance is more than the width of usual road.In order to realize nothing People drives, it is necessary to realizes the much more accurate determination of vehicle location.

Preferably, it is used as starting point using the optimal presumed value of vehicle location 501.This for example can be the position that GNSS is determined Put.It can also be the newest vehicle location known.The information be forwarded to described comprising home environment 504 vehicle database 253.Home environment description can include the data relevant with road, such as, but not limited to road equipment and/or geographical environment.Should Environmental sensor data 502 of the data with that may have been obtained by environmental sensor 210 preferably in data associating unit closes Connection 505.It is highly preferred that the position sensor data 503 that may have been obtained by vehicle location sensor 230 can be used for further Association.Associated by the way that home environment at least is described into 504 with environmental sensor data 502, the vehicle location of renewal can be obtained Information 506.

In fig. 4 it is shown that the data flow of road characteristic data difference detection.Again, vehicle location preferably can be used 501 optimal presumed value is as starting point.This for example can be GNSS defined locations.It can also be the newest vehicle position known Put or the vehicle location of renewal from step 506.The position is forwarded to vehicle database 253, and it is preferably at least transmitted Road characteristic data 513, road equipment data 514 and environmental data 515.In addition, ring can be obtained by environmental sensor 210 Border sensing data 512.Preferably, environmental sensor data passes through specific data processing or image procossing 523 to generate data Data needed for associated steps.The road carried out using the road characteristic data 513 and environmental sensor data 512 of vehicle database Road performance data association 516 preferably generates the data for including road characteristic data difference 519.Utilize the road of vehicle database The road equipment data correlation 517 that facility data 514 and environmental sensor data 512 are carried out is preferably generated comprising road equipment The data of data differences 520.In addition, the ring carried out using the environmental data 515 and environmental sensor data 512 of vehicle database Border data correlation 516 preferably generates the data for including environmental data difference 521.The difference data of generation can include and depend on inspection The more or less information for the difference amount measured.Situation for not detecting difference, difference can not include further letter Breath or the mark that further information can not be only obtained comprising instruction.For such case, vehicle database 253 does not transmit any Specific data, such as road data 513, road equipment data 514 or environmental data 515, difference can be included by environmental sensor The complete data set that data 512 provide.Or environmental sensor data passes through the condensing routine in data compressor or computer Further compression.After being serviced device database request, vehicle can provide complete data set or even additional data, such as Image or parts of images.

In Figure 5, acquirement and the road object of environmental data collection are illustrated in greater detail together with Fig. 6 a, 6b and 6c Generation/assessment.In general, term environmental data collection is used for the data set generated by multiple sensors herein, it is described more Individual sensor may include at least one optical image sensor and/or radar sensor and/or infrared sensor and/or other rings Border sensor.According to the combination and/or association of these sensor signals, the environmental data collection from vehicle visual angle is collected, according to The environmental data collection generates one group of road object.Vehicle 110 travels on road 300.The road have can by boundary line, Kerbstone or any other border or the right margin 310 and left margin 311 of indicia meanses mark.In existing at the center of road Heart line, it is dotted line in the present embodiment, and the center line has a plurality of line 320,330,340,350 and 360.Every line With beginning and end.Starting point is marked as 321,331,341 and 351, and terminal is marked as 322,332,342 and 352. The center line of any other type may be present or center line even not be present.Additionally, there are some road equipments, and it can be handed over Logical mark 370,371,372.According to the environmental data collection of capture, vehicle preferably identifies above-mentioned road object and most preferably Generate corresponding second data set.

Although environmental data collection can provide 360 degree of expressions of vehicle environmental, shown here by referring to observation fan section 380 The more limited visual field, observation fan section 380 may correspond to preceding video camera, and it has about 90 ° of viewing angle.Corresponding front Image is illustrated in Fig. 6 a.Here a part for road object, i.e. right margin 310 and traffic only can be seen on the right side of image Mark 380.Show center line line 330 on the left of image and center line then rule 340 a part.Obviously, this is regarded Angle gives the picture rich in detail of the related road object of right-hand lane of the road just travelled wherein to vehicle, but can not provide The excessively information relevant with the road object of part thereof of left-hand lane can be seen in the upper left corner of image.When vehicle is along road When road is advanced through position 111 and 112, it is according to the capture images of image fan area 381 and 382.Corresponding image is together with identification The road object gone out is illustrated in Fig. 6 b and 6c together.

Match the continuous representation that the serial road object will provide road.Can not by this matching with from panoramic camera Or the image mosaic that panorama software obtains is compared.Wherein, it is only necessary to identify the match flag in adjacent image and necessary phase Should ground zoomed image.In order to match road object, it is necessary to drive the space conversion of associated object data collection.This conversion can To be linear displacement, scaling or even matching the complicated non-linear conversion of the data of capture.It will illustrate later this The details of conversion.These conversions partly will be facilitated in itself by road object, and this can allow sensor in the second data set and Find to carry out automatic recalibration during consistent error between detected road object in one data set.In fig. 7 it is shown that Acquirement and assessment for the driving associated object data collection of system-wide section.As illustrated by Fig. 5, first group of environmental data 380th, 381 and 382 captured by the first vehicle, and be converted into and drive associated object data collection, first vehicle is in the figure In advance in the first direction from right to left in right-hand lane on track above.Second group of environmental data is by the second vehicle Capture, and be converted into and drive associated object data collection, second vehicle is in the figure on the left side on the track of lower section Advanced in reverse direction from left to right on the track of side.By the second vehicle generate driving associated object data collection can be it is similar, As shown in Fig. 6 a, 6b and 6c.Main difference is that driving associated object data collection now mainly comprises the road on second lane Therefore object simultaneously provides the information relevant with second lane.So far, also no vehicle has complete one group of road of system-wide section Road object, but server database the information provided by vehicle can be fitted together with obtain provided so far it is complete One group of whole road object.Therefore, vehicle generation is rule with each road object such as right margin 310, left margin 311, center line 320th, 330,340,350,360 and the relevant object-based information of traffic sign 370,371,372.The object-based information Collected by server database and assessed, with second of the two-step pretreatment for generating and updating accurate road characteristic data storehouse The complete driving associated object data collection of road is generated in object data processing step.This example is not limited to two vehicles, but can Extend to any number of vehicle.Furthermore, it is not required that single unit vehicle provides the continuous data stream for driving associated object data collection.Example Such as, the object-based information related to first environment data set as shown in Figure 6 a can be provided by the first vehicle, and Fig. 6 b The object-based information related to second environment data set can be provided by the second vehicle later by same road, and be schemed 6c the 3rd environmental data collection can be captured by the 3rd vehicle for again passing by same road later.

In fig. 8 it is shown that the example of object packet.The packet of object can be used for by vehicle to adjacent environmental data collection Or simplified and preferably matching to driving associated object data collection and providing by vehicle or server database.Substantially, vehicle can Packet is used for the first object data processing step.Most preferably, server database will be grouped and be used at the second object data Manage step.

A large amount of relative position benchmark, such as center line line may be present along road.Due to substantially being looked in the presence of many Similar dotted line, therefore, it is difficult to make a distinction and obtain absolute position benchmark between each dotted line.Anyway, line can be used for Determine position of the vehicle relative to road.In order to verify and correct GNSS location information, it is necessary to more absolute positions benchmark.It is such Position reference can seldom be occurred or at least only occur object once in GNSS tolerance separations.These benchmark can be handed over Logical mark, irrigation canals and ditches, manhole or the only characteristic indication on road.They can be grouped with pavement marker such as dashed centre line Together.Include center line line 350 and traffic sign 372 for first group 401.Include center line line 330 for second group 402 and hand over Logical mark 370 and the traffic sign 371 positioned at the opposite side of road.By referring to these object groups, driving phase can be calculated Close object data set or the correct conversion with the group or at least part thereof of other sensing data.This conversion can be with It is linear displacement, scaling or even matching the complicated non-linear conversion of captured driving associated object data collection. This allows the more preferable matching of the driving associated object data collection shot from road, even if the three-dimension varying of captured image or Error be present in the measurement parameter for being generated by it the driving associated object data collection of vehicle.This improves the object mould generated The precision of type.For example, if the driving associated object data collection for the second vehicle for being shown as group 401, traffic sign can be obtained 372 by the farthest corner of the corresponding driving associated object data collection in the vehicle travelled in reverse direction, but it is corresponding in The line of heart line can also be identified well by the vehicle travelled in reverse direction.This is schematically depicted in more detail in the next figs.

In figures 9 a and 9b, the first example matched by being grouped object is shown.Fig. 9 a are shown positioned at Fig. 5's The view of the vehicle travelled from left to right on the track of lower section.Here, traffic sign 372 can be clearly seen and can be with Center line line 350 is grouped together to form the first group objects 401.Because traffic sign is in main view Yezhong, so can To be accurately determined its position.In addition it is also possible to it is accurately determined the relative position of center line line 350.Fig. 9 b show position In the view of the vehicle travelled from right to left on the track of top.Traffic sign 372 is only located at the farthest corner of image simultaneously And apart from far.The pattern such as triangle formed by the object and/or proportioning that match packet by the object being grouped, it is possible to achieve The easy matching of the driving associated object data collection shot by the different vehicle travelled along different directions.Can be by determining object Characteristic point and/or edge come improve object packet and matching.In the present embodiment, the He of first end 351 of center line line 350 The characteristic pattern that second end 352 is used to generate the first group objects 401 together with the pedestal of traffic sign 372.Of the pattern It is accurate complete to generate with allowing each driving associated object data collection fragment and/or object's position Accurate Curve-fitting together Whole object factory.As shown in figures 9 a and 9b, due to substantially having recognized identical triangle pattern, thus can easily by Two drive associated object data collection and/or object's position matching together, although they are obtained from different visual angles.

In addition, the matching determines available for exact position.The vehicle travelled from right to left in Figure 5 can not use traffic mark Will 372, because the traffic sign 372 is in the most peripheral in the visual field of the vehicle.By the center line for identifying the first group objects 401 Line 350, can be compared accurate positioning.Based on the predetermined information of vehicle by travelling from left to right, along anti- The vehicle of direction running knows the exact position of the center line line 350 associated with traffic sign 372.

Figure 10 a and 10b show the second example matched by being grouped object.Here, show and drawn including center line Second group objects 402 of line 330 and traffic sign 371.The profile of the group by center line line 330 first end 331 and second The pedestal of the pedestal and traffic sign 371 of end 332 and traffic sign 370 limits.In figure loa, show by along Fig. 5 road The first of the vehicle generation that road up travels from right to left on track drives associated object data collection.Here, it illustrate only group 402 part.In figure 10b, the different driving related objects shot by different vehicle and/or different cameras are shown Data set.The driving associated object data collection uses the imaging sensor capture with large viewing and therefore displays that traffic mark Will 370.Here, all members of the second group objects 402 are shown.It can will at least include center line line 330 and traffic mark First section of the group 402 of will 371 associates with previous driving associated object data collection.By referring to positioned at road opposite side Traffic sign 371, the further proportioning of the object with going out by the vehicle detection travelled in reverse direction can be carried out.

In fig. 11 it is shown that Data Collection, distribution and the basic procedure of processing.Specifically, the first object in vehicle The combination of data processing step 555 and the second object data processing step 559 in server database causes with high accuracy With the data set of accurate positional information.According to this method, bag is updated using the vehicle to be communicated with server database Include the accurate road characteristic data storehouse for driving related object.The figure has two row.Left column shows the action related to vehicle, and Right row show the action related to server database.Server database preferably memory of driving related object number of objects According to.It is this driving associated object data may further comprise at least positional information and with road object, road equipment object, geographic object The relevant object-based details of the other object related to driving.

In the first step, vehicle for example based on programme path or interest region from server database solicited message 550. Or server database can trigger transmission 551, such as to update customizing messages.After this request or triggering, server 552 at least one first data set related to request or triggering of database generation.These first data sets are based on being stored in clothes The 4th data set being engaged in device database.Preferably, each first data set includes driving associated object data.Preferably, take Business device database has been stored with the information relevant with the state of vehicle database.This information may include to be stored in the number in vehicle According to the backup of collection or at least one identifier for identifying the data set being stored in vehicle.The status information is preferably in car Have determined that and to update after the reception of the first data set.By using the status information, the first data set can only include modification Data set and/or new data set.Data set, the vehicle that vehicle does not have, which may be present, to be had but updated data set, vehicle The data set that the data set or vehicle for having but being retransmitted should be deleted.Therefore, if asked by vehicle, described at least one Individual first data set can include the data relevant with desired overlay area.If triggered by server database, it is described extremely Few first data set can include the data updated and/or alarm data and/or the data that should be verified by vehicle.It is preferred that Ground, each first data set include at least one positional information and detailed object-based information.Positional information is specified detailed Object-based information position, region or the area that can be applied to.

In the next step, the first data set is forwarded to vehicle 553 and be stored in vehicle database.Next, Vehicle passes through at least one sensor collection environmental data 554 as described above.The collection can be along at least one spy in its path Determine section and/or occur at specific time and/or specific object.Preferably, it is related right to be related to driving for environmental data collection As at least one of, road equipment and geographic object.This environmental data is located in the first object data processing step 555 Reason, the second data set of object-based information is included with generation.The processing may include to be based on environmental data and/or the first data At least one statistical estimation concentrated.It can further comprise the packet, matching and conversion of data as described above.In addition Step 556 in, the 3rd and the first difference between data set are preferably calculated, to generate comprising the section phase with vehicle route 3rd number of the difference between the object-based information of the second data set and the object-based information of the first data set closed According to collection.In addition, at least one 3rd data set may include to replace at least one second data set with the difference of the first data set Object-based information.If necessary to and/or being serviced device database request, then these the 3rd data sets can also include new number According to and/or independently of existing first data set data.In next step 557, the 3rd data set is forwarded to service Device database.Server database stores the 3rd data set 558 received and opened in the second object data processing step 559 Begin to handle them.It is this to handle the statistical estimation for preferably including the 3rd data set and further post processing.In addition, it can be used The 4th data set being stored in server database.This post processing may include that object packet, object matching, group are matched, turned Change, count calculating and other processing.Then, the 4th data set can be updated 560 using its result.

In fig. 12 it is shown that the same part accurately matched by vehicle with road of the part as step 555 The detail flowchart of related vehicle route data.In step 570, vehicle based at least one second data set come identify with The same part correlation of road and the multiple first vehicle route sections to overlap each other at least in part.Then, in step 571 In, multiple object datum marks are identified in each vehicle route section.Datum mark is preferably what can easily and precisely be identified Point.Datum mark be grouped into step 572 so that, at least two vehicle route sections of multiple first vehicle route sections In, each group includes identical object datum mark.Later, generate or calculate in step 573 for matching and/or matching and road At least one set of object at least two vehicle route sections of multiple first vehicle route sections of the same part correlation on road The proportioning function of datum mark.This proportioning function may include times in linear translation and/or rotation or even more complicated conversion One.All other second data set application identical of the vehicle route section of identical at least two is matched somebody with somebody in step 574 Than function, to generate or update the 4th data set for including driving associated object data.

Generation match function after, can by preferably calculate the 3rd or second data set average value and/or standard Deviation statistically handles the second data set.

In fig. 13 it is shown that the part as step 559 is accurately matched and road by server database With the detail flowchart of the related vehicle route data of a part.In step 580, server database is based at least one the Three data sets identify multiple first vehicle route areas that are related to a same part for road and overlapping each other at least in part Section.Then, in step 581, multiple object datum marks are identified in each vehicle route section.Datum mark is preferably to hold The point identified easily and accurately.Datum mark be grouped into step 582 so that, in multiple first vehicle route sections at least In two vehicle route sections, each group includes identical object datum mark.Later, generate or calculate in step 583 and be used for Matching and/or at least two vehicle route sections for matching the multiple first vehicle route sections related to a same part for road In at least one set of object datum mark proportioning function.This proportioning function may include linear translation and/or rotation or even more Any one of complicated conversion.To all other 3rd number of the vehicle route section of identical at least two in step 584 According to collection using identical proportioning function, to generate or update the 4th data set for including driving associated object data.

, can be by preferably calculating the average value and/or standard of the 4th or the 3rd data set after generation matches function Deviation statistically handles the 3rd data set.

In fig. 14 it is shown that pinpoint detail flowchart.Using the data from database and by vehicle on edge Collected data carry out being accurately positioned for vehicle during one route.In first step 590, vehicle receives from database First data set, first data set is based on the 4th data set including driving associated object data.This driving is related right Image data can comprise at least positional information, optional object datum mark and drive related object such as road pair with least one As, road equipment object, geographic object or the relevant object-based details of the other object related with driving.Work as row When sailing, in step 591, vehicle generates the multiple of multiple sections at least along its path by least one sensor of vehicle Environmental data collection.Environmental data collection, which is related to, drives at least one of related object, road equipment and geographic object.In step In 592, vehicle drives associated object data by assessing at least one environmental data collection and being generated based on assessment result At least one second data set is handled to perform object data.In step 593, during traveling and the second data are being generated After collection, vehicle attempts to identify identical driving related object included in the first data set and/or right in the second data set As datum mark and/or multigroup driving related object and/or object datum mark.In step 594, vehicle match the second data set with Identical between first data set drives related object and/or object datum mark and/or multigroup driving related object and/or right As datum mark, and in step 595 calculate or derive the positional value of its own position.

Figure 15 show total image processing load in vehicle is mitigated and expand it is increasingly automated drive it is (HAD) and other Basic 2 stage method of transportation database is established while the use range of wagon control application.Figure 11 step 553 it Afterwards, vehicle receives the first data set in step 600.In comparison step 601, by the confidence levels of each first data set with Limiting value is compared.If confidence levels over-limit condition, start second stage, i.e. stage 2 to the first data set.It is no Then, the first data set incipient stage 1 is utilized.

Figure 16 illustrates in greater detail the stage 1.Here, collect other environmental data in step 610, as in Fig. 4 more It is illustrated in detail in.Later, off-line data processing is started in step 611 using environmental data.Although it is preferred that for example exist After driving or when not sampled to data for being travelled on the section that further handles collected by processed offline Environmental data, but also handled online in the case where sufficient disposal ability can be obtained.In step 612, select Treatment Options.This Treatment Options have hereinbefore been disclosed more closely in and may include：

Scaling 613：The environmental data collection close to object and farther environmental data collection are selected, to carry out the object phase Multiple step format triangulation and positioning for other objects.This has obtained good identification using lean Processing Algorithm and accurately determined Position result.

Feature 614：Selection is concentrated close to the environmental data collection of object of interest in each environmental data.

Move back and forth 615：Move back and forth environmental data collection so as to from closely or with different view and apart from determination interest Object.

Sequence 616：A series of environmental data collection is run successively by different algorithm groups.

Look back 617：Retroversion processing is carried out from nearest environmental data collection to past environmental data collection.

After one or more of these options are selected, the 3rd data set is generated in step 618.

Figure 17 illustrates in greater detail the stage 2.First, environmental data is collected in step 620.In step 621, selection The method that related object is driven for the second data set generation.Step 620 and 621 commutative orders.Therefore, environment number can be directed to According to type or other parameters select this method or data can be collected on demand by this method.Decision to method is based primarily upon The type of first data set.If the first data set includes the object factory of the driving related object on airway road, first is selected Method.If the first data set includes allowing by different way and showing as with different view the Vector Graphics of image, Select second method.

If system of selection 1, handle the object factory of the first data set.If system of selection 2, is carried out to image level Other comparison.It disclosed above the details of this method.By using any means, counted based on the first data set generation 624 second According to collection.In addition, two methods can be used to same data set.

Except shown here as the step of in addition to, vehicle it is detectable and confirm the first data set.It can be collected newspapers via the 3rd data Accuse difference.

The selection of method can be carried out, so as to start to collect environmental data.In such a case it is not necessary to select step 621.

Figure 18 shows the image of the exemplary roadway with multiple driving related objects.

Figure 19 shows the driving related object in previous image.Road 701-704, street sign indicator 710-717, red green be present Lamp 720 and 721 and traffic sign 730-733.The first data set being storable in vehicle preferably includes driving dependency number Object data.This driving associated object data may further comprise at least positional information, optional object datum mark and with least one It is individual to drive related object such as road object, road equipment object, geographic object or relevant with the other object of driving correlation The data of object-based details and the Vector Graphics including allowing to present as image.First data set can enter One step includes confidence level information.

Figure 20 is shown according to the object related to driving and including first number related preferably with the object of previous accompanying drawing The image presented according to the data of the vector graphics format of collection.These images presented can be used for letter of the generation for wagon control Breath, and/or the information for vehicle location, and/or for by by least one environmental data collection in image level and figure Picture or parts of images are compared to generate and update the second data set of the first data set.Environmental data collection can be in advance along it Path is by least one sensor collection of vehicle, and environmental data collection is preferably with driving related object, road equipment and ground Manage at least one of object correlation.Or can be by by least one environmental data collection and the driving for including airway road The object factory of at least one first data set of the object factory of related object is compared to generate information.

Claims

1. a kind of be used to carry out using the data received from database and the data collected when along a route by vehicle The method of accurate vehicle positioning, this method comprise the following steps：

A) vehicle is received including for the current of the vehicle from the database and with object-based 3D vector formats First data set of the driving associated object data of position, this driving associated object data comprise at least location information, optional Object datum mark and with least one driving related object such as road object, road equipment object, geographic object or with driving The relevant object-based details of the other object of correlation are sailed,

B) at least one sensor that the vehicle passes through the vehicle while driving along at least multiple sections in its path Multiple environmental data collection are generated, the environmental data collection is with driving at least one in related object, road equipment and geographic object Person is related,

C) vehicle is by assessing at least one environmental data collection and based on assessment result with object-based 3D vector formats At least one second data set that generation drives associated object data is handled to perform object data,

D) when driving and after the second data set of generation, the vehicle attempts the number of identification first in second data set According to identical the driving related object and/or object datum mark included in collection and/or multigroup driving related object and/or multigroup right As datum mark, and

E) vehicle matches identical between the second data set and the first data set using object-based 3D vector formats Driving related object and/or object datum mark and/or multigroup driving related object and/or more group objects datum marks, to derive The positional value of its own position.

2. according to the method for claim 1,

Characterized in that,

Institute's location value includes and the skew for the value measured by the GNSS system of the vehicle.

3. according to the method for claim 1,

Characterized in that,

Institute's location value is included in the position world of first data set relative to included in first data set Drive related object and/or multigroup position for driving related object.

4. according to any method of the preceding claims, it is characterised in that

The vehicle is simplified and/or accelerated object handles and/or the life of second data set using first data set Into, and/or to identical the driving related object and/or object benchmark included in first data set in the second data set Point and/or multigroup identification for driving related object and/or more group objects datum marks.

5. according to any method of the preceding claims, it is characterised in that

The vehicle utilizes imaging sensor, infrared sensor, laser scanner, ultrasonic transducer, radar sensor and RF letters At least one of road sensor collects at least one environmental data collection, and described image sensor is, for example, ccd sensor.

6. according to any method of the preceding claims, it is characterised in that

The database is by least one static trust server positioned at the outside vehicle.

7. according to any method of the preceding claims, it is characterised in that

At least one set of object includes at least one of road object and road equipment object and geographic object.

8. according to any method of the preceding claims, it is characterised in that

Driving at least one second data set of related object includes at least location information of the GNSS based on vehicle.

9. a kind of vehicle, it has the device for being used for that accurate vehicle positioning to be carried out using the data from database, the data Stock contain include drive associated object data data set, it is this driving associated object data comprise at least positional information, can The object datum mark of choosing and with it is at least one driving related object for example road object, road equipment object, geographic object or with The relevant object-based details of related other object are driven,

The vehicle comes from the number with the current location for being used to receive with object-based 3D vector formats for vehicle According to the device of first data set in storehouse,

The vehicle, which has, to be used to pass through at least one of the vehicle along at least multiple sections in its path while driving Sensor generates the device of at least one environmental data collection, and at least one environmental data collection is with driving related object, road Facility is related at least one of geographic object,

For by assessing at least one environmental data collection and being given birth to based on assessment result with object-based 3D vector formats The device of object data processing is performed at least one second data set for driving associated object data,

For identifying the device of multiple object datum marks at least one second data set,

For generating the device of at least one set of object datum mark at least one second data set,

Described in attempting when driving and being identified after the second data set of generation at least one second data set Identical driving related object and/or object datum mark included in first data set and/or multigroup driving related object and/or The device of more group objects datum marks, and

For using object-based 3D vector formats at least one second data set and first data set it Between match identical and drive related object and/or object datum mark and/or multigroup driving related object and/or more group objects benchmark The device of point and the positional value for deriving its own position.

10. vehicle according to claim 9,

Characterized in that,

At least one sensor of the vehicle is ccd sensor, infrared sensor, laser scanner, ultrasonic transducer, radar Sensor or RF channel sensors, for collecting at least one environmental data collection.